More useful might be to post the actual price. The Hagibis MC40 is ALWAYS on a huge discount. Typical pricing is around US$55-65, but is occasionally lower than that and occasionally higher than that.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Thunderbolt 3 M.2 NVMe SSD Enclosures
- Thread starter oreggero
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
Sorry, sick at the moment. I think this should convert to less than $40 https://www.aliexpress.com/item/1005006484705885.html
Just got my hagibis usb 4 enclosure in combination with the kingston renegade 2tb nvme ssd.
Boy, this thing uses close to to 5w in idle which is pretty insane. 2w in sleep.
as far as temps goes iStat Menus is telling me that the ssd temp is 35 celsius which is pretty good.
With the SSD and HDD attached my mac mini uses 13w in idle which is really non ideal IMO. Need to figure out a way to wake the mini from sleep efficiently. Regular WOL magic packets do not wake my mini properly - it goes back to sleep after 30 seconds.
edit: around 25-28 celsius during sleep, slowly ramps up to 35 during idle. pretty good
Boy, this thing uses close to to 5w in idle which is pretty insane. 2w in sleep.
as far as temps goes iStat Menus is telling me that the ssd temp is 35 celsius which is pretty good.
With the SSD and HDD attached my mac mini uses 13w in idle which is really non ideal IMO. Need to figure out a way to wake the mini from sleep efficiently. Regular WOL magic packets do not wake my mini properly - it goes back to sleep after 30 seconds.
edit: around 25-28 celsius during sleep, slowly ramps up to 35 during idle. pretty good
Last edited:
How about a "fast enough" enclosure that sips power ? I'd read many of these run pretty hot and use > 1w even with the machine at idle.
I'd read the enclosures using the Intel chipset are slightly slower but use far less power. Maybe that's a good compromise ?
I'd read the enclosures using the Intel chipset are slightly slower but use far less power. Maybe that's a good compromise ?
Interesting. When connected through my Plugable Thunderbolt 4 / USB 4 hub, my Kioxia XG8 + Hagibis MC40 is identified as a Thunderbolt 3 device.
This is contrast to my Samsung 990 Pro + Qwiizlab ES40UR connected directly to the M4 Mac mini. It is identified as a USB 4 device, as was the Kioxia / Habigis when it was connected directly to the Mac mini. I don't know what this means, if anything.
EDIT: I just plugged the Samsung / Qwiizlab into the hub, and now it is identified as a Thunderbolt 3 device too.
Furthermore, the device revision and the vendor ID both change.
1. That graph you posted is efficiency, not power utilization. Often drives that are faster are more efficient, but might actually use significantly more power at peak. IOW, they may have high bursts of speed, but at higher power, whereas slow drives may not be very efficient, but use less power at peak.
2. Efficient high performance drives are bottlenecked by the PCIe 3.0 interface of these USB 4 enclosures, meaning those efficiency numbers are meaningless because they cannot achieve their high performance to generate that high efficiency.
3. Efficient high performance drives may use higher power not only at peak but also at more moderate speeds. So, if such a drive is bottlenecked by the enclosure, it may become even lower efficiency than other so-called lower efficiency drives.
4. Some of those drives are DRAM-less, which often use less power than drives with DRAM. The efficiency measurements are with Windows machines that can make use of HMB for DRAM-less drives. That efficiency is lost on Macs and in external enclosures, because there is no HMB support.
5. ASM2464PD enclosures use significant power with any drives. Many so-called efficient drives + these USB 4 enclosures may use 5 - 5.5 Watts combined at idle.
Real world example: I own the Kioxia XG8 4 TB in a Hagibis MC40, and the Samsung 990 Pro 4 TB in the identical Qwiizlab ES40UR enclosure. Both have the exact same ASM2464PD chip with the exact same 47.3 firmware. The Kioxia XG8 runs much cooler than the Samsung 990 Pro despite the Kioxia's much lower power efficiency rating:
Both are drives with DRAM. Performance in Black Magic and AmorphousDiskMark is in the same ballpark, mainly because the drives are bottlenecked by the enclosure. In Windows testing, the 990 Pro is much, much faster but in USB 4 enclosures, they are neck and neck in those tests.
This is contrast to my Samsung 990 Pro + Qwiizlab ES40UR connected directly to the M4 Mac mini. It is identified as a USB 4 device, as was the Kioxia / Habigis when it was connected directly to the Mac mini. I don't know what this means, if anything.
EDIT: I just plugged the Samsung / Qwiizlab into the hub, and now it is identified as a Thunderbolt 3 device too.
Furthermore, the device revision and the vendor ID both change.
Yes and no. A few things to consider:
1. That graph you posted is efficiency, not power utilization. Often drives that are faster are more efficient, but might actually use significantly more power at peak. IOW, they may have high bursts of speed, but at higher power, whereas slow drives may not be very efficient, but use less power at peak.
2. Efficient high performance drives are bottlenecked by the PCIe 3.0 interface of these USB 4 enclosures, meaning those efficiency numbers are meaningless because they cannot achieve their high performance to generate that high efficiency.
3. Efficient high performance drives may use higher power not only at peak but also at more moderate speeds. So, if such a drive is bottlenecked by the enclosure, it may become even lower efficiency than other so-called lower efficiency drives.
4. Some of those drives are DRAM-less, which often use less power than drives with DRAM. The efficiency measurements are with Windows machines that can make use of HMB for DRAM-less drives. That efficiency is lost on Macs and in external enclosures, because there is no HMB support.
5. ASM2464PD enclosures use significant power with any drives. Many so-called efficient drives + these USB 4 enclosures may use 5 - 5.5 Watts combined at idle.
Real world example: I own the Kioxia XG8 4 TB in a Hagibis MC40, and the Samsung 990 Pro 4 TB in the identical Qwiizlab ES40UR enclosure. Both have the exact same ASM2464PD chip with the exact same 47.3 firmware. The Kioxia XG8 runs much cooler than the Samsung 990 Pro despite the Kioxia's much lower power efficiency rating:
Both are drives with DRAM. Performance in Black Magic and AmorphousDiskMark is in the same ballpark, mainly because the drives are bottlenecked by the enclosure. In Windows testing, the 990 Pro is much, much faster but in USB 4 enclosures, they are neck and neck in those tests.
Last edited:
Of course NVMe can saturate a Thunderbolt 4 (TB4) connection, Thunderbolt 5 enclosures are coming.
And HMB is well known to be not supported by Apple. https://medium.com/@a23e/m-2-ssds-for-your-mac-75bb3635fdb5
DRAM-less of course uses less power.
I've used a ASM2464PD TB4 enclosure since early days (Zike drive) with a single-sided 4TB DRAM-less NVMe drive (Lexar). It does run quite hot, though updating the firmware helped some.
The Lexar used to be quite cheap, not so cheap anymore on Amazon.
I just bought a slow 8TB drive for backup use.
I'm waiting for NVMe 8TB drives to become cheap, then I'll move the Lexar to a slow enclosure that's much more portable than the Zike, and use the 8TB NVMe drive in the Zike.
I think at this point since you have a TB5 capable machine, why not wait until the TB5 portable enclosures come out (and then come down a bit in price)?
Back when I got the Zike drive, it was nearly the only game in town. If I had waited, I would have saved some money, but I needed the 4TB external storage to move stuff off my 4TB Apple SSD. Also, the Lexar 4TB was not a crazy price, currently at a ridiculous $313.46. The Zike drive still runs hot, 56ºC according to smartctl, but it's below its limits of course.
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Well, this is even more interesting. 😳
As an experiment, I moved my 990 Pro + Qwiizlab to a Thunderbolt 4 / USB 4 hub. It is no longer directly connected to my Mac mini. Idle power utilization has dropped from 5.0-5.5 Watts (plugged into Mac mini directly) to 3.1-4.6 Watts (connected through hub). I also measured the XG8 + Hagibis, and its power utilization in this setup connected through a hub is 4.1 Watts. Upper is 990 Pro / Qwiizlab, lower is XG8 / Hagibis:
Yes, I have the luxury of testing the two drives through two identical TB 4 / USB 4 hubs in parallel. In the pic below, it's the original orientation, upper XG8 / Hagibis, and lower 990 Pro / Qwiizlab. Note that the Qwiizlab is reading 3.6 W idle here:
Again, as mentioned in my prior post, note that when connected through these Thunderbolt 4 / USB 4 hubs, these are detected as Thunderbolt 3 drives, and not USB 4 drives. It would seem that in Thunderbolt mode, idle power utilization is decreased vs USB 4. With this setup, the idle temps of the two drives are much closer, with the XG8 at 37 C and the 990 Pro at 38 C. When the 990 Pro was attached directly to the Mac mini, it idled at about 41 C. (Room temp about 20 C.) That means that switching from USB 4 mode (directly connected to Mac mini) to Thunderbolt 3 mode (connected through a hub) dropped the idle temp by 3 C.
In this setup, both drives have no sleep issues at all, and there are no disconnect issues either.
I don't have pure Thunderbolt enclosures to test, but my tests above shine a new light on the ASM2464PD enclosures.
Last edited:
^^^ After running these two drives in this configuration - through TB 4 / USB 4 hubs - over two days, I can confirm without a doubt that Thunderbolt 3 mode runs cooler than USB 4 mode with the same drives. I wonder if the only way to force Thunderbolt 3 mode with these enclosures is to use such a hub. I can't see how to force Thunderbolt 3 mode otherwise.
With this extended monitoring, I can also confirm that in this configuration connecting through hubs, both drives usually idle at 3.6-4.1 W. (Shortly after taking that 4.6 W picture, it settled to 4.1 W or less, so maybe it was doing something else when I took that picture.) This represents about a 1-2 Watt drop at idle, compared to when the drives are directly connected to the Mac mini (5.0-5.5 W). For this reason and because of the lower idle temperature, and because I've encountered zero glitches, I will leave the drives in this configuration going forward.
BTW, both hubs are connected to the Mac mini with true Thunderbolt 4 cables. However, the cables I'm currently using to the connect the drives to the hubs are both USB 4 cables, but that doesn't seem to matter. Regardless if I use a certified Thunderbolt 4 cable or a USB 4 cable to connect the drive directly to the Mac mini, the drive gets detected as a USB 4 device. OTOH, regardless if I use a certified Thunderbolt 4 cable or a USB 4 cable to connect the drive to a hub, the drive gets detected as a Thunderbolt 3 device.
With this extended monitoring, I can also confirm that in this configuration connecting through hubs, both drives usually idle at 3.6-4.1 W. (Shortly after taking that 4.6 W picture, it settled to 4.1 W or less, so maybe it was doing something else when I took that picture.) This represents about a 1-2 Watt drop at idle, compared to when the drives are directly connected to the Mac mini (5.0-5.5 W). For this reason and because of the lower idle temperature, and because I've encountered zero glitches, I will leave the drives in this configuration going forward.
BTW, both hubs are connected to the Mac mini with true Thunderbolt 4 cables. However, the cables I'm currently using to the connect the drives to the hubs are both USB 4 cables, but that doesn't seem to matter. Regardless if I use a certified Thunderbolt 4 cable or a USB 4 cable to connect the drive directly to the Mac mini, the drive gets detected as a USB 4 device. OTOH, regardless if I use a certified Thunderbolt 4 cable or a USB 4 cable to connect the drive to a hub, the drive gets detected as a Thunderbolt 3 device.
Thanks for your testing, this is interesting. Let's see if we can find a way to force TB3/TB4 instead of USB4 without hubs.
A question about Hagibis MC40:
They told me the product does not ship with the square blue thermal pad for the ASM2464PD chip. So I was thinking of buying a suitable thermal pad myself. But I couldn't find any information on the optimal thickness, should I go for something like 3 mm or is that too much
They told me the product does not ship with the square blue thermal pad for the ASM2464PD chip. So I was thinking of buying a suitable thermal pad myself. But I couldn't find any information on the optimal thickness, should I go for something like 3 mm or is that too much
Mine came with 2 thermal pads for the chip and 2 for the ssd
Thanks for letting me know!
Their own demonstration video includes them as well, it's just that when I contacted Hagibis on Aliexpress they told me only SSD thermal pads are included. Do you know how thick the chip pads are?
Same here, both for the Hagibis and the Qwiizlab.
According to my ruler, the little square is 4 mm thick. Also, the ones I got were not blue. For both my Hagibis and Qwiizlab, unlike in that video they were grey pads with a blue cover (with the cover to be removed at installation), and are likely high conductivity thermal pads.
Last edited:
Intel JHL7440-controlled enclosures are explicitly Thunderbolt 3; ie do not behave as if USB4, where USB4 differs from TB3
I have 2x such enclosures connected to ports of a TB4 nano hub; & yes they are recognised/listed as TB3
Last edited:
I would like to get an OWC Express 4M2, but I have a few (stupid) questions.
Primarily I'm concerned about storing RAW photos. In the beginning I would not use all 4 slots, but only 2.
1. From what I understand, if all slots are not occupied, the storage is not operating at the highest possible speed?
2. I would like to go the RAID 1 way, i.e.: mirroring the disks. Do I need to have purchased SoftRAID? I have the option to buy storage without a SoftRAID XT license (about $120) cheaper, but I don't know if there is a free alternative. I certainly won't use all the advanced features that SoftRAID offers, I just need to throw something on a disk and know that it exists there mirrored. And that SoftRAID XT license that comes with these repositories is a lifetime license? I don't think I mind the $120, but I don't want to pay the extra monthly fees.
Primarily I'm concerned about storing RAW photos. In the beginning I would not use all 4 slots, but only 2.
1. From what I understand, if all slots are not occupied, the storage is not operating at the highest possible speed?
2. I would like to go the RAID 1 way, i.e.: mirroring the disks. Do I need to have purchased SoftRAID? I have the option to buy storage without a SoftRAID XT license (about $120) cheaper, but I don't know if there is a free alternative. I certainly won't use all the advanced features that SoftRAID offers, I just need to throw something on a disk and know that it exists there mirrored. And that SoftRAID XT license that comes with these repositories is a lifetime license? I don't think I mind the $120, but I don't want to pay the extra monthly fees.
You may get a free period of SoftRAID with a new drive - I certainly did with all mine. Although saying that, mine were pre-populated. You can always use Apple Raid instead if you prefer - many have used this without issue also, although I personally stick to SoftRAID.
You only need the paid version of SoftRAID to create the RAID array (which is why I think you get a period free when you get a new drive). Otherwise the free version of SoftRAID will be fine for monitoring and using the drive etc.
Bear in mind the 4M2 has a built in fan - it can be quite noisy. Better to get the Thunderblade, which are completely silent with no moving parts - but cost a lot more too......!
I have no experience with that OWC enclosure, but generally speaking, any multi-NVMe solution that isn't in the form of a PCIe card, is a waste of time with too many restrictions and problems.
Software RAID1 can be achieved by just Apple Disk Utility, though without a hardware RAID controller, the restorability when one disk does fail is in question (as with any software RAID anyway).
Since you are not stripping with RAID0 to create a single large and fast volume, I think you may need to sit back a step and explore if this is the solution you want. I am struggling to think of a real world scenario where a RAID1 mirror of bus powered NVMe is the best solution for.
Didn't OWC just announce Thunderbolt 5 Thunderblade? They claim 6,500 MB/s.
So, naturally, the worth of the older generation is lowered.